Antenatal Glucocorticoid Exposure and Long-Term Alterations in Aortic Function and Glucose Metabolism
OBJECTIVE: Animal studies have demonstrated long-term effects of in utero glucocortcoid exposure on vascular development and glucose metabolism. We hypothesized that there would be a similar impact in humans.
METHODS: One hundred and two young adults born preterm aged 23 to 28 years, with prospective data collection from birth, and 95 adults born term after uncomplicated pregnancies underwent cardiovascular MRI. We compared cardiac and aortic structure and function, as well as cardiovascular risk profile, in a nested case-control study of 16 participants exposed to antenatal steroids and 32 who were not, but with otherwise similar perinatal care. Outcomes were compared with normal ranges in those born term.
RESULTS: Adults whose mothers had received antenatal steroids had decreased ascending aortic distensibility (9.88 ± 3.21 vs 13.62 ± 3.88 mm Hg−1 × 103, P = .002) and increased aortic arch pulse wave velocity (5.45 ± 1.41 vs 4.47 ± 0.91 m/s, P = .006). The increase in stiffness was equivalent to that of term adults a decade older. Those who had in utero exposure to antenatal steroids also had significant differences in homeostatic model assessments for β-cell function (P = .010), but in multiple regression analysis this did not explain the impact of steroids on aortic function.
CONCLUSIONS: Antenatal glucocorticoid exposure in preterm infants is associated with increased aortic arch stiffness and altered glucose metabolism in early adulthood.
- AA —
- ascending aorta
- CI —
- confidence interval
- DA —
- distal descending aorta
- ECG —
- HOMA —
- homeostasis model assessment
- LV —
- left ventricular
- PDA —
- proximal descending aorta
- PWV —
- pulse wave velocity
What’s Known on This Subject:
In utero exposure to glucocorticoids in animal models influences vascular development. Studies in young adults have shown that exposure to antenatal glucocorticoids alters glucose metabolism, but it is not known whether there are any cardiovascular effects.
What This Study Adds:
Glucocorticoid exposure is associated with a localized increase in aortic arch stiffness, similar in magnitude to term-born individuals a decade older. The change in stiffness does not relate to changes in glucose metabolism that were also evident in this cohort.
Glucocorticoids are potent regulators of growth and development and, in animal studies, transient glucocorticoid exposure during critical windows of fetal life results in long-term changes in elastin synthesis,1 vascular function,2–5 baroreceptor function,6 sympathetic outflow, hypothalamic-pituitary-adrenal axis, nephrogenesis, renin-angiotensin system, and glucose homeostasis.7 Whether in utero exposure to glucocorticoids has similar long-term effects in humans has, traditionally, been difficult to study experimentally. However, preterm birth is a unique clinical scenario in which a range of interventions, including antenatal steroids, are administered during a critical developmental period.8–12 We have undertaken phenotyping of a large cohort of young people born preterm, with detailed prospective perinatal data collection, to investigate for evidence of developmental programming in humans, and define the biological impact of different exposures.13 Elastin deposition within the aortic wall is critically altered in animal models by glucocorticoid exposure in late gestation.14 Therefore, we have used this cohort to test the specific hypothesis that in utero glucocorticoid exposure in humans alters aortic structure. Furthermore, we have determined the relevance of alterations in glucose metabolism, previously observed in preterm infants exposed to antenatal steroids,15 to this association and whether the altered aortic function has had an impact on cardiac structure and function. Finally, we related the degree of aortic stiffening to that observed during aging in the healthy population to gauge the relative impact of antenatal steroid exposure on long-term cardiovascular risk for those born preterm.
We performed a 25-year follow-up of a cohort of preterm-born individuals from 5 centers in the United Kingdom, who were randomly assigned to feeding regimes between 1982 and 1985.16 The initial cohort consisted of 926 subjects (birth weight <1850 g) with obstetric and neonatal details recorded at birth.13 Follow-up was performed at 18 months,17 7 years,18 and 15 years19 of age. When they had reached the age of 23 to 28 years, we invited 240 who had agreed to be contacted about future studies to attend an appointment for detailed cardiovascular phenotyping (Fig 1). We recruited adults aged 20 to 39 years born term after uncomplicated pregnancies to undergo the identical investigations for comparative, age-stratified, normal ranges of our outcome measures. The study was approved by the relevant ethics committee (Oxfordshire Research Ethics Committee A: 06/Q1604/118), and all participants provided signed informed consent.
Demographics, Metabolic Characteristics, and Blood Pressure
Subjects attended in the morning after a 12-hour overnight fast. Height was measured to the nearest centimeter, and weight was measured to the nearest 0.2 kg, with subjects wearing light clothing, by using a combined digital height and weight measurement station (Seca, UK). Data on medical history, smoking, parental medical history, and lifestyle were obtained by using a validated questionnaire.20,21 Blood samples were drawn, centrifuged, and separated within 30 minutes, then stored for later analysis at −80°C. Fasting blood biochemistry was measured at the Oxford John Radcliffe Hospital Biochemistry Laboratory by using routine validated assays, with clinical level quality controls. Low-density lipoprotein cholesterol was calculated by using the Friedewald formula.22 Homeostatis model assessments for insulin sensitivity (HOMA-S) and β-cell function (HOMA-B) were calculated based on fasting insulin and glucose levels by using a HOMA calculator (http://www.dtu.ox.ac.uk/homacalculator/index.php).
Three blood pressure measurements were recorded for each participant on the left arm with an automatic digital monitor (HEM-705CP, OMRON, Japan), and the second and third measurements averaged for analysis.16 Central blood pressure was assessed based on left radial artery applanation tonometry to derive ascending aortic pressure waveforms (SphygmoCor Analysis System, Australia).16
Cardiac Magnetic Resonance
Aortic Pulse Wave Velocity
Aortic pulse wave velocity (PWV) was measured by cardiovascular magnetic resonance on a 1.5 Tesla (Siemens, Sonata) scanner.23 Images were captured by using a free breathing, retrospectively electrocardiogram (ECG)-gated, spoiled gradient echo sequence. Velocity-encoding gradient for phase contrast was applied to measure through-plane flow in the ascending aorta (AA) and proximal descending aorta (PDA) at the level of the pulmonary artery and then for the distal descending aorta (DA) 11 cm down from this. Image analysis was performed by using analytical software (Argus, Siemens Medical Solutions, Germany). Flow images were manually contoured as previously described.23 Aortic PWV was calculated as the distance between the assessed sites of the aorta (measured on an aortic oblique sagittal image, Fig 2) divided by the time delay between the arrival of the foot of the pulse wave at each site, determined as the intersection of the tangent of the upslope of the flow curve with the x-(time) axis. Global PWV was calculated from AA to DA and for the aortic arch (AA to PDA) to provide information on regional stiffness.23
Variation in stiffness in different sections of the aorta was determined independently by measurement of aortic distensibility in the same positions as for PWV. ECG-gated, steady-state free precession cine images were acquired during breath hold. Maximum and minimum aortic cross-sectional areas over the cardiac cycle were determined by using semiautomated edge detection algorithms (Matlab, Mathworks Inc), and distensibility was calculated as the relative change in area divided by pulse pressure.24 Pulse pressure was derived by using SphygmoCor central blood pressure measures (SphygmoCor Analysis System, Australia).
Images were acquired prospectively ECG-gated during end-expiration breath hold by using a steady-state free precession left ventricular (LV) short-axis stack to determine LV volumes, ejection fraction, and mass index for body surface area, calculated by using Argus (version B15, Siemens Healthcare, Germany).
Statistical analysis was performed by using version 18 of SPSS (SPSS Inc, Chicago, IL). We matched each case exposed to antenatal steroids with 2 controls not exposed, but with similar gestational age, birth weight z-score, gender, key maternal risk factors, pregnancy complications, and postnatal complications to limit confounding (Table 1). In older populations, an increase of 0.8 m/s in PWV equates to a 10% increased risk of total cardiovascular events.25 We identified 16 participants exposed to antenatal steroids and determined that matching 1 case to 2 controls would provide 80% power, at P = .05, to identify a 0.74 SD difference between groups (equivalent to a 0.8 m/s difference in PWV) and 2.6 mm Hg−1 × 10−3 difference in AA distensibility based on normal ranges for these variables in young populations.13
Normality of variables was assessed by visual assessment of normality curves and Shapiro–Wilk test. PWV was log transformed for normal distribution. Comparison between groups for continuous variables was performed by using a Student’s t test or 1-way analysis of variance for normally distributed data and Mann–Whitney and Kruskal–Wallis tests for skewed data. For categorical variables, comparison was done by using a χ2 test.
Stepwise linear and hierarchical regression models, with AA distensibility or arch PWV as the dependant variable, were used to determine the variables independently predicting aortic arch stiffness within this cohort. Spearman bivariate correlations were performed to identify variables achieving correlation coefficient P values <0.1 for inclusion in regression models. Within sets of interrelated variables, such as blood pressure or glucose homeostasis measures, the variable with strongest correlation was included. Blood pressure variables were excluded from the distensibility model because this is corrected for in the calculation of distensibility. Results are presented as mean ± SD. P values <0.05 were considered statistically significant.
The mothers of 16 participants received steroids and, because antenatal glucocorticoid use varied significantly between recruitment centers (Supplemental Table 3), consistent with local practice in the 1980s, we were able to match these individuals to 32 participants whose mothers had not received steroids, in a nested case-control design. Matching was performed based on the variables listed in Table 1. There were no differences between groups in feeding regimes in the neonatal period or weight gain in the first year of life (results not shown). In early adult life there were no significant differences between groups in age, family cardiovascular medical history, personal medical history, or lifestyle (results not shown). Our control group of 95 individuals consisted of 39 (41%) males, had an average age of 28 years (range, 20–39 years), mean birth weight of 3365.20 ± 494.52 g, and mean gestational age of 39.59 ± 0.92 weeks.
Antenatal Steroids and Cardiovascular Risk Factors
There were no significant differences in anthropometry, blood pressure, LV, or lipid profile measures between preterm-born individuals exposed or not exposed to antenatal steroids (Table 2).
There was a significant reduction in β-cell function (HOMA-B) in those exposed to antenatal steroids (P = .02) (Fig 3C), driven by lower insulin levels (P = .04) and higher glucose levels (P = .08) (Fig 3 A and B). Because information on fasting glucose and insulin was available for a proportion of subjects (7 cases and 15 controls) from the follow-up at age 15 years,19 we also studied whether similar trends were evident in adolescence. The difference in glucose was significant in adolescence (P = .03) with a difference in HOMA-B of the same magnitude, although this did not reach significance (P = .2). At this age, levels of insulin were similar. There were no differences in insulin sensitivity at either age point (HOMA-S) (Fig 3D).
Antenatal Steroids and Arterial Stiffness
We then studied whether antenatal steroid exposure was associated with variation in aortic stiffness. There was no significant difference in global PWV (P = .25), although PWV was 1 m/s faster in the arch of the aorta in those exposed to steroids (P = .006) (Fig 4A). This regional increase in aortic stiffness was confirmed by distensibility measures, because there was a significant reduction in AA distensibility (9.88 ± 3.21 vs 13.62 ± 3.88 mm Hg−1 × 10−3, P = .002) and preserved levels more distally along the aorta (Fig 4B). These results were not explained by variation in aortic geometry (Table 2). We also looked at the flow profiles at the different sections of the aorta by averaging the individual data for each group separately to ensure there were no differences in the upslope gradient that may have modified our ability to accurately identify the intercept. The gradient and shape of the curves was the same in both groups with the only difference being the earlier arrival of the pulse wave at the PDA in those exposed to antenatal steroids, consistent with our calculated PWV measures (Fig 2).
Antenatal Steroids and Aortic Aging
We then compared the levels of aortic stiffness in our preterm offspring with those observed in individuals born term. In the normal population, as age increases there is a proportional increase in aortic stiffness in the AA and aortic arch.26 We therefore plotted age against AA distensibility for the 95 subjects born term after uncomplicated pregnancies aged 20 to 39 years and demonstrated a significant linear relation (B = −0.483 mm Hg−1/year, 95% confidence interval [CI] −0.884 to −0.405, r2 = 0.232, P < .001) (Fig 4C). Based on this standard association, we used the mean and SEM of AA distensibility observed in the 2 preterm-born groups to interpolate “aortic ages.” For those not exposed to steroids, it was 22.19 ± 1.9 years, similar to the actual mean age of the preterm cohort. For those exposed to in utero steroids it was nearly 10 years greater than their chronological age (32.55 ± 2.22 years) (Fig 4C).
Glucose Homeostasis, Aortic Stiffness, and Antenatal Steroids
Glucose homeostasis variables were correlated with both AA distensibility and arch PWV (HOMA-B and arch PWV, r = −0.336, P = .03; AA distensibility, r = 0.307, P = .05) in our preterm cohort. We therefore performed regression analyses to determine the independence of the association between antenatal steroids and aortic stiffness from other factors.
Six variables achieved a correlation with AA distensibility with a P value <0.1 and were included in the regression model (Supplemental Table 4). By use of stepwise regression, antenatal steroid exposure was an independent predictor of AA distensibility (β = −.440, 95% CI −5.927 to −1.551, P = .001). In hierarchical multiple regression, gender, gestational age at delivery, Apgar score at 5 minutes, HOMA-B, and high-density lipoprotein accounted for 46.7% of the variance, and inclusion of steroid exposure to this model accounted for an additional 13.4% of the variance in AA distensibility (r2 change = 0.134, F change = 11.053, P = .002).
With the use of the same method for aortic arch PWV, 4 independent variables were included (Supplemental Table 4). Antenatal steroid exposure was an independent predictor of aortic arch PWV (β = .295, 95% CI 0.005–0.119, P = .034). We therefore concluded that exposure to antenatal steroids was an independent determinant of aortic stiffness in young individuals.
Our study demonstrates for the first time that young people born preterm, whose mothers received antenatal steroids, have a regional increase in aortic stiffness in the AA and aortic arch. Furthermore, they have changes in glucose metabolism that are independent of the impact of antenatal steroids on aortic function.
Aortic stiffness predicts future development of hypertension and is an independent predictor of coronary artery disease, stroke, and mortality.26 In older adults, a 1 m/s increase in PWV relates to a 14% increase in total cardiovascular events.25 There are no studies of aortic stiffness in young populations with long enough follow-up to provide outcome data. However, increased availability of imaging techniques such as cardiovascular magnetic resonance has demonstrated that aging during the first decades of life is characterized by increases in ascending aortic stiffness.26,27 We also demonstrated this phenomenon in our term-born study group and used these data to determine that preterm-born individuals whose mothers received antenatal steroids had levels of aortic stiffness equivalent to adults in their thirties.
Changes in vascular structural matrix proteins are a key determinant of vascular aging28 and the third trimester, when antenatal glucocorticoids are given in human pregnancy, is the period of most rapid aortic remodeling during life.29 At this time, there is accelerated arterial elastin accumulation and cross-linking,14,30 resulting in an acute increase in aortic wall stiffness and PWV.29,31 This change in aortic stiffness is thought to be a preadaptation to the dramatic increase in newborn blood flow after cessation of placental blood flow and initiation of pulmonary gas exchange. In animals, antenatal glucocorticoids modified this maturation process and increased the elastin precursor, tropoelastin, in the fetal aorta,14 and enhanced expression of lysyl oxidases, critical in cross-linking elastin and collagen, as well.32 However, Doppler studies showed profound increases in fetal blood flow sustained for up to 96 hours after glucocorticoids.33 Differential flow exposure during development could lead to disparity in aortic geometry or upper-body vessel formation, which would complicate comparisons of aortic stiffness.34 However, there was no variation in aortic flow patterns or geometry in our cohort, and, because increased flow also drives elastin accumulation, our observations would be consistent with specific effects of glucocorticoids on elastin maturation. The histologic structure of the aorta varies along its length, allowing it to function as a reservoir and conductive vessel to turn pulsatile flow into laminar flow,35 and the AA is richest in elastin.36,37 We have previously shown that postnatal intravenous lipid exposure has a selective impact on the distal aorta. This region is preferentially associated with lipid accumulation in the arterial wall in neonatal life13 and, together, these observations highlight the value of detailed cardiovascular phenotyping to disentangle complex interactions between interventions and aortic development.
Infants whose mothers received antenatal glucocorticoids have a lower incidence of hypotension and decreased need for pressor support postnatally,38,39 consistent with a steroid-induced accelerated maturation of aortic structure. A key advantage of this maturation is that it optimizes central hemodynamic function and preserves myocardial and cerebral perfusion pressure in preterm-born neonates. In the current study, we found no difference in LV measures related to whether subjects were exposed to antenatal glucocorticoids. However, once these individuals reach adult life, the degree of increase in aortic stiffness observed is likely to have an adverse cardiovascular effect, and longer-term cardiovascular follow-up in this cohort will be of interest.
Those exposed to antenatal glucocorticoids had a specific reduction in β-cell function. Follow-up of adults aged 30 years whose mothers took part in a randomized controlled trial of antenatal betamethasone15 also demonstrated differences in glucose metabolism in response to an oral glucose tolerance test, with higher plasma insulin levels in those exposed to steroids. We did not perform oral glucose tolerance tests but replicated our findings based on samples collected 10 years earlier. The mean gestational age in our cohort was 29 weeks, whereas, in the previous study, ∼30% were born term. The impact of glucocorticoid exposure on glucose metabolism may vary with gestational period, because glucocorticoid signaling is most relevant to pancreatic islet formation earlier in fetal development.40 Further work will be able to define whether effects vary depending on dose and time of glucocorticoid administration.
A significant strength of our study is the prospective collection of maternal and perinatal data, which allowed us to identify a precisely defined cohort of preterm-born individuals exposed to antenatal steroids. Although our study is the largest prospective follow-up study of preterm infants with this level of detail on cardiovascular phenotyping, the proportion exposed to steroids is relatively small and they were not randomly assigned. However, at the time of recruitment, use varied significantly between centers. We were therefore able to match cases to controls with equivalent perinatal care, which would not be possible in younger cohorts because of widespread use of antenatal steroids. In the late 1980s and throughout the 1990s, postnatal glucocorticoids were also used in the United Kingdom to treat preterm infants with bronchopulmonary dysplasia.41 It is possible that postnatal glucocorticoid exposure may also have detrimental effects on aortic development. However, our participants were recruited before this period, and thus we are unable to assess the impact of this intervention in our cohort.
It is possible that steroid use has led to survival of infants who would not have survived otherwise, resulting in a cohort with more severe aortic dysfunction. However, the findings remain of clinical relevance to the long-term cardiovascular health management of those exposed to antenatal steroids. Interestingly, 30% to 50% of steroid doses are currently given to women who go on to deliver after 34 weeks11,15 and, therefore, whose infant gains no specific benefit from the therapy. It will be of interest to understand whether this cohort also has differences in aortic function.
Antenatal glucocorticoids for women at risk for preterm delivery are the standard of obstetric care and reduce postnatal premature infant morbidity. Between 1975 and 2000, the percentage of preterm infants exposed to antenatal steroids rose from 8% to 75%.10 We demonstrate that, 25 years later, in utero steroid exposure is associated with changes in glucose metabolism and localized changes in aortic function in young people. Although antenatal glucocorticoids irrefutably improve the survival of preterm-born infants, the impact on the cardiovascular system may need to be considered in the longer-term care of this population when they reach adulthood.
- Accepted December 23, 2011.
- Address correspondence to Paul Leeson, PhD, MRCP, Oxford Cardiovascular Clinical Research Facility, Department of Cardiovascular Medicine, University of Oxford, John Radcliffe Hospital, Oxford, OX3 9DU United Kingdom. E-mail:
All authors made substantial intellectual contribution to the design of the study and writing of the manuscript. The final version was approved by all authors.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Supported by grants from the Oxford Health Services Research Committee, National Institute for Health Research Oxford Biomedical Research Centre, and Oxford British Heart Foundation Centre for Research Excellence. Dr Leeson is supported by the British Heart Foundation (FS/06/024 and FS/11/65/28865), and Mr Lewandowski is supported by the Commonwealth Scholarship Commission.
- Swee MH,
- Parks WC,
- Pierce RA
- Lewandowski AJ,
- Lazdam M,
- Davis E,
- et al
- Lazdam M,
- de la Horra A,
- Pitcher A,
- et al
- Leeson CPM,
- Kattenhorn M,
- Morley R,
- Lucas A,
- Deanfield JE
- Friedewald WT,
- Levy RI,
- Fredrickson DS
- Redheuil A,
- Yu W-C,
- Wu CO,
- et al
- Bendeck MP,
- Langille BL
- Chinoy MR,
- Zgleszewski SE,
- Cilley RE,
- Krummel TM
- Moïse AA,
- Wearden ME,
- Kozinetz CA,
- Gest AL,
- Welty SE,
- Hansen TN
- Copyright © 2012 by the American Academy of Pediatrics